Not so. The AI was always there. We originally intended to use Text-to-Speech for output but rapidly found that TTS couldn't deliver the vocal nuances and timin

Not so. The AI was always there. We originally intended to use Text-to-Speech for output but rapidly found that TTS couldn't deliver the vocal nuances and timing required for the experience we wanted to deliver - this after all was a game where humour and quality of writing were crucial - hence the move to pre-recorded chunks. One of the heroic and oft-unsung latter-day development efforts on the game was the design of the pre-recorded chunks for assembly on the fly at play time and the mechanism by which they were assembled and delivered.